Background: Little research has examined how anchor numbers affect choice, despite several decades of research showing that judgments typically and robustly assimilate toward irrelevant anchors.
Methods: In one experiment, HIV-positive patients (N = 99) judged the chances that sexual partners would become infected with HIV after sex using a defective condom and then indicated their choices of remedial action. In a second experiment, Iowa physicians (N =191) rated the chances that hypothetical patients had a pulmonary embolism and then formulated a treatment plan.
Results: Irrelevant anchor numbers dramatically affected judgments by HIV-infected patients of the chances of HIV infection after a condom broke during sex (43% v. 64% in the low- and high-anchor conditions, respectively) and judgments by doctors of the chances of pulmonary embolism (23% v. 53%, respectively). Despite large anchoring effects in judgement, treatment choices did not differ between low-and high-anchor conditions. Accountability did not reduce the anchoring bias in the doctors' judgments.
Discussion: The practical implications of anchoring for risk judgments are potentially large, but the bias may be less relevant to treatment choices. The findings suggest that the theoretical underpinnings of the anchoring bias may be more complex than previously thought.